Proposition 3.3 (third claim) — strict exactness implies (3.4): minimizers of the relaxation have and
ProvedRockafellianRelax.proposition_3_3_coptimizationp2o-gran-per-chapterp2o-plan-paperp2o-v1stochastic-optimizationvariational-analysis
Let be a Rockafellian for the problem of minimizing . If is strictly exact, supported by , then for all ,
Under strict exactness, solving the relaxation solves the actual problem.
Differences from the source:
- encoding: ℝᵐ × ℝⁿ, values in R̄ = [−∞, ∞], ⟨y, u⟩ → (Fin m → ℝ) × (Fin n → ℝ) with the product topology, EReal, y ⬝ᵥ u; reason: Lean convention; same instances (the product topology has the same convergent sequences as the Euclidean norm)
Preamble
import Mathlib import Definitions.Def_RockafellianRelax_IsStrictlyExact import Definitions.Def_RockafellianRelax_argmin import Definitions.Def_RockafellianRelax_tilt
Formal statement
namespace RockafellianRelax
/-- **Proposition 3.3, third claim: strict exactness gives (3.4).** Royset, Chen & Eckstrand,
arXiv:2204.04762v4, Proposition 3.3, pp.7–8: "Suppose f : ℝᵐ×ℝⁿ → R̄ is a Rockafellian for the
problem (3.1) of minimizing φ : ℝⁿ → R̄. … If f is strictly exact, supported by ȳ, then
(u⋆, x⋆) ∈ argmin_{u,x} f(u, x) − ⟨ȳ, u⟩ =⇒ u⋆ = 0, x⋆ ∈ argmin_x φ(x). (3.4)"
**Formalization Note.** `IsStrictlyExact f φ ȳ` includes that `f` is a Rockafellian for `φ`. -/
theorem proposition_3_3_c {m n : ℕ} (f : (Fin m → ℝ) × (Fin n → ℝ) → EReal)
(φ : (Fin n → ℝ) → EReal) (ybar : Fin m → ℝ) (hsx : IsStrictlyExact f φ ybar) :
∀ (us : Fin m → ℝ) (xs : Fin n → ℝ), (us, xs) ∈ argmin (tilt f ybar) →
us = 0 ∧ xs ∈ argmin φ := by sorry
end RockafellianRelax
Source
Royset, Chen & Eckstrand, Rockafellian Relaxation and Stochastic Optimization under Perturbations, arXiv:2204.04762v4, Proposition 3.3, pp.7–8, (3.4)
Human review
Confirmed by the mission captain (proposal self-audit).
Confirmed by the moderator at approval.